Bypass caps for LM4562?

KanedaK

Member
2011-02-07 8:32 pm
I've bought a couple LM4562 to replace the NE5532 in my ReVox B226.

In the original circuit there aren't any bypass caps on the opamps supply pins.

I have ceramic multilayer axial 0.01uF at hand; can I use those? I see more often people use 0.1uF but sometimes 0.01uF is considered better?

What would happen if I parallel a bigger electrolytic with one of those 0.01uF? ( i see a little bit of everything when it comes to decoupling opamps)

Also, what about a cap between pin 4 and pin 8? instead of the two others? Or should I put three (on v+, on v-, and between)?
 

Mooly

Administrator
Paid Member
2007-09-15 8:14 am
You'll probably find some decoupling within a centimetre or so of the opamp. I'd be surprised if there were none.

A small (=<10uf) electrolytic in series with a 1 or 2.2 ohm 0.25 watt connected across pins 4 and 8 can be as good as anything for general situations.
 
It is true, in B226 there are no local filter capacitors near the supply legs of IC101 and IC102. On the other hand, IC201/202 are supported by 100uF on each supply leg but they have a different purpose (input stage of headphone amplifier).
As in one of your previous topics, I recommend using the conventional 0.1uF MLCC between each supply leg and ground. The 0.01uF value you mention works in the 100MHz range which is not relevant neither for the noise in this CD player nor for the op amps you intend to use.
Adding capacity anywhere in the analogue section of B226 in my experience kills the dynamics and so does a capacitor across the supply pins of op amps.
 

KanedaK

Member
2011-02-07 8:32 pm
It is true, in B226 there are no local filter capacitors near the supply legs of IC101 and IC102. On the other hand, IC201/202 are supported by 100uF on each supply leg but they have a different purpose (input stage of headphone amplifier).
As in one of your previous topics, I recommend using the conventional 0.1uF MLCC between each supply leg and ground. The 0.01uF value you mention works in the 100MHz range which is not relevant neither for the noise in this CD player nor for the op amps you intend to use.
Adding capacity anywhere in the analogue section of B226 in my experience kills the dynamics and so does a capacitor across the supply pins of op amps.

0,1uF have been ordered and should be delivered soon.
That said, my b226 just sounds better and better each day, result of all the new caps settling in / breakin in, and I almost feel reluctant to change the opamps (desoldering and soldering looks more difficult than for capacitors, and the Revox PCB is supr fragile...)

Is it really worth going for the LM4562? What kind of improvement can one excpect on this particular player?
 

KanedaK

Member
2011-02-07 8:32 pm
It is worth it. Less digital and more analogue, smoother sound, less nervousness and graininess, more low-level detail, better bass definition, better soundstage.
But of course, it is possible that it will work the other way around for you. :)

Haha OK. Less graininess and more low-level detail sounds cool. Better soundstage as well (not that it is "bad" now).
I just don't wanna lose attack and "jump factor" wich are now just how I like them to be :) overly smooth I find boring. I don't wanna turn my revox into an Audio Analogue :)

Maybe I should start by buying a better, controlled temperature soldering iron.
 

KanedaK

Member
2011-02-07 8:32 pm
It is worth it. Less digital and more analogue, smoother sound, less nervousness and graininess, more low-level detail, better bass definition, better soundstage.
But of course, it is possible that it will work the other way around for you. :)

I just finished installing the LM4562.
0,1uF between each supply leg and ground.

It's only been playing a few songs but I'm not totally convinced of the result. Somehow, nothing "bad" jumps to the ears, but it also lacks a little bit sharpness and conviction. I wonder if I bypassed them correctly. Doesnt sound unbalanced tho, nor harsh, just a bit too "recessed" and "soft". maybe i need to get used to the sound, or / and maybe they need a few hours burn-in.
Also there seems to be a bit less gain than with NE5532.

I will let them play for a while and see if things "improve".
 
I think I wrote the caveat in one of my replies that in your case changes might have adverse effects. :)
One change at a time is the advice you should always keep in mind: you should have listened to new op amps without bypassing first. Anyway, bypassing does not introduce detrimental effects to the sound if done properly.
I cannot tell about LM4562, my recommendation is LME49720 which I have played with a lot and which brings to B226 exactly the changes you seem to be missing now. Anyway, an op amp will require several hours to open up the sound.
 

KanedaK

Member
2011-02-07 8:32 pm
I think I wrote the caveat in one of my replies that in your case changes might have adverse effects. :)
One change at a time is the advice you should always keep in mind: you should have listened to new op amps without bypassing first. Anyway, bypassing does not introduce detrimental effects to the sound if done properly.
I cannot tell about LM4562, my recommendation is LME49720 which I have played with a lot and which brings to B226 exactly the changes you seem to be missing now. Anyway, an op amp will require several hours to open up the sound.

I guess LME49720 and LM4562 are supposed to be the same part. Some pretend they sound slightly different, but being electrically the same, I would expect if LME49720 needs bypass caps, so does LM4562, their behaviour cannot be that different. Also I figured B226 having no local supply bypass at all, it wasn't worth trying without the caps :)

That said, music keeps playing as I'm writing this and doing things around the house, and it seems (it SEEMS ) it's already a bit more balanced and open. Fresh cold opamps cannot sound at their best, in my experience nothing else in audio sounds good when cold and new...
 

Windforce85

Member
2014-11-08 11:59 am
Just pair of 10 or 22uF aluminium-electrolytic caps no further than 10mm from both supply pins to ground will be the best decoupling for the LM4562. Don't need to use ceramics, it can do much more harm to the sound than any benefits. If it's oscillating or noisy this will be problem with ground loops. Solid ground plane below LM4562 all the way is strongly recommended.
 

KanedaK

Member
2011-02-07 8:32 pm
Just pair of 10 or 22uF aluminium-electrolytic caps no further than 10mm from both supply pins to ground will be the best decoupling for the LM4562. Don't need to use ceramics, it can do much more harm to the sound than any benefits. If it's oscillating or noisy this will be problem with ground loops. Solid ground plane below LM4562 all the way is strongly recommended.

I used 0,1uF MLCC as recommended by Willi Studer (and as seen on many other threads and forums).
So far, after two hours of playing, the sound is becoming very good and is a major improvement over the NE5532. Much more detail, less distortion >> less metallic high-mids, more extended treble, bass goes lower and is more defined. Stereo image is more airy and extends a bit wider on the sides. I guess the shut-in and hollow sound I experienced on the first songs played were due to a brand new cold IC, not to oscillation. I will of course listen more before any definitive judgement, but I like what I hear now.
 
As expected. You could even try to cover most of the PCB tracks with capacitors' ends to ensure minimum inductance. It makes a difference, in fact it is easily noticeable.

The next step could be to try to place the 10uF low ESR electrolytic 2cm away from Vcc / Vee pins.... the exact placement will depend on PCB tracks width and thickness. The resulting sound should be very pleasant analog-like to listen to, and very detailed with correct placement of instruments in 2D / 3D space in front of you.

Nick
 

KanedaK

Member
2011-02-07 8:32 pm
As expected. You could even try to cover most of the PCB tracks with capacitors' ends to ensure minimum inductance. It makes a difference, in fact it is easily noticeable.

The next step could be to try to place the 10uF low ESR electrolytic 2cm away from Vcc / Vee pins.... the exact placement will depend on PCB tracks width and thickness. The resulting sound should be very pleasant analog-like to listen to, and very detailed with correct placement of instruments in 2D / 3D space in front of you.

Nick

Would 10uF Panasonic FR series ultra-low ESR caps be a good choice there?
 

KanedaK

Member
2011-02-07 8:32 pm
As expected. You could even try to cover most of the PCB tracks with capacitors' ends to ensure minimum inductance. It makes a difference, in fact it is easily noticeable.

The next step could be to try to place the 10uF low ESR electrolytic 2cm away from Vcc / Vee pins.... the exact placement will depend on PCB tracks width and thickness. The resulting sound should be very pleasant analog-like to listen to, and very detailed with correct placement of instruments in 2D / 3D space in front of you.

Nick

Would you have a picture of that disposition maybe?
 

KanedaK

Member
2011-02-07 8:32 pm
I'm not completely satisfied with the sound.
I'm afraid my bypassing might not be as effective as it sould.
In the Revox B226, in the original circuit there's only 22uF on each rail, few inches away from the opamps, just after the regulators.

All I did was adding 0,1uF between each opamp supply pin and ground, but my opamps are in sockets (machined and gold-plated) wich I read isn't the best since it effectively lenghtens the distance between the opamp and the decoupling capacitor by about 5mm.

The problem is, this Revox pcb is super fragile and no matter how careful you are, it's easy to lift pads. So I figured, if I have good sockets and good decoupling, once that is fixed, I can swap opamps easily without risking to damage the pcb by soldering and desoldering again and again...

Now I can't tell if the LM4562 are oscillating?... I don't think they are? They're, at east, not warmer to the touch than the NE5532 that are still in there in the variable outputs circuit; there's no lack of bass, quite the opposite, there's more bass now than before, it's just some lack of body and solidity in the midrange, for the rest it's all better, but there was some presence in the midrange that was very pleasing with the NE5532 that is now gone.

On the other hand, I believe in burn-in, I've experienced it many times, I have a very very revealing system, and these opamps have about 25/30 hours of playing... I have read in few different places that LM4562 mellows down and gets body after about 100 hours, but I know burn-in and capacitor sound is a very controversial subject specially on this website...

Anyways, I'm collecting as much info as I can, seeing pictures of good implementations would help (a good picture is worth a thousand words) but there aren't so many modded revox players around here I guess... In the meantime, I'll wait for the "100 hours" before deciding if I should mod further or not.
 

KanedaK

Member
2011-02-07 8:32 pm
This morning I tried 100uF Panasonic FC in parallel with 0,1uF MLCC from each supply pin to ground.
Warmer, but a bit slow and muddy, plus I could "hear" the Pana FC cap instead of the Silmic II that is on the rail just after the regulator.

I'm now trying just 0,1uF MLCC between pin 4 and pin 8, combined with another 0,1uF from V- to ground.

So far this gives the most natural sound, I will listen a bit further.